 The novel JKS DGCN algorithm was developed for the purpose of accurately classifying sleep stages based on multiple biosignals. It uses the standard CNN to extract features from each signal, then uses two adaptive adjacency matrices to connect signals from the same EPIC or nearby EPICs. This allows the model to capture the transition between sleep stages. Additionally, the model employs a jumping knowledge spatial temporal graph convolution module to extract spatial features from the graph convolutions and temporal features from standard convolutions. This enables the model to learn the transition rules between sleep stages. Finally, the model's execution speed is evaluated using the ISRIC-S3 and ISRIC-S1 data sets, where it achieves the best performance compared to other models. This article was authored by Xiaoping Ji, Yan Li, and Peng Wen.